Medical Report Classification on MIDRC
98.8Micro PrecisionBioClinicalBERT
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| BioClinicalBERTTraining Mode=Baseline2026.04 | 98.8 | 70.6 | 82.4 | |
| Qwen-7BTraining Mode=SFT + GRPO2026.04 | 98.1 | 89.3 | 93.5 | |
| LlamaTraining Mode=SFT2026.04 | 98 | 82.4 | 89.5 | |
| MedalpacaTraining Mode=SFT + GRPO2026.04 | 97.2 | 87.6 | 92.1 | |
| LlamaTraining Mode=SFT + GRPO2026.04 | 95.1 | 98.3 | 96.7 | |
| MedalpacaTraining Mode=SFT2026.04 | 92.2 | 90.1 | 91.1 | |
| Qwen-7BTraining Mode=SFT2026.04 | 91.9 | 95.8 | 93.8 | |
| Qwen-3BTraining Mode=SFT2026.04 | 82.7 | 56.3 | 67 | |
| Qwen-3BTraining Mode=SFT + GRPO2026.04 | 76.7 | 58 | 66.1 | |
| GPT4-oTraining Mode=Baseline2026.04 | 66.7 | 98.4 | 79.5 | |
| Phi3Training Mode=SFT2026.04 | 58.2 | 48.7 | 50.9 | |
| Phi3Training Mode=SFT + GRPO2026.04 | 56 | 51.3 | 53.5 | |
| Qwen-7BTraining Mode=Baseline2026.04 | 32.7 | 21 | 25.5 | |
| Qwen-3BTraining Mode=Baseline2026.04 | 29.8 | 14.3 | 19.3 | |
| LlamaTraining Mode=Baseline2026.04 | 25.4 | 14.3 | 18.3 | |
| Phi3Training Mode=Baseline2026.04 | 24.3 | 15.1 | 18.6 | |
| MedalpacaTraining Mode=Baseline2026.04 | 0 | 0 | 0 |